基于改进集合预测网络的输变电设备故障知识图谱构建方法

doi:10.19595/j.cnki.1000-6753.tces.241362

Abstract
Figure/Table
References
Related Citation (2)

Download: PDF (0 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract With the intelligent transformation and upgrading of power grids, higher requirements have been put forward for the analysis and maintenance of power system operation status. As an important part of the power grid, power transmission and transformation equipment may cause unpredictable losses in case of failure. Knowledge Graph describes concepts, entities and their relationships in the objective world in a structured form, providing a tool for better organizing, managing and understanding information from different sources and structures, so the introduction of knowledge in transmission and transformation equipment management and troubleshooting is conducive for professionals to quickly locate fault information and make effective decisions.
To effectively integrate the scattered knowledge related to power transmission and transformation and improve the quality of knowledge graph construction of power transmission and transformation equipment faults, this paper carried out the knowledge graph construction based on the relevant specification manuals of power transmission and transformation, and further improves the entity-relationship extraction algorithm. Existing entity-relationship extraction methods in the power field ignore the connection between triples and have problems of weak text representation, vague entity localization, and low accuracy of classification of long-tail relationships. In response to the above, this paper proposed a joint entity-relationship extraction model based on the improved set prediction network (ISPN). The set prediction network (SPN) is used to model the overall triples, which introduces unsupervisSed contrast learning to BERT encoder for enhanced text representation. The enhanced text representation is then fed into a decoder to guide triple queries to learn potential triple features in the text. The triple queries contain global and boundary features to capture information at different scales. The prediction triples are obtained by the feed forward neural network. In order to improve the performance of entity boundary identification, this paper proposes to use boundary regression algorithm to model the boundary offset of entities, which uses decoder output features to predict the start and end index offset of head and tail entities firstly, then the candidate triples are obtained according to the binary matching algorithm, finally correct the entity boundaries according to the boundary regression formulae, and also use the boundary loss instead of the cross-entropy loss in total loss. In the relationship classification stage, cost-sensitive learning is introduced to balance the losses of different types of triples, and the classification accuracy of long-tail relationships is improved by assigning larger loss weights to long-tail relationships as well as those with high misclassification rates to facilitate the model to effectively learn the features of long-tailed relationships. The redesigned binary matching loss is used to model the difference between the predicted set and the true set and optimize the model training.
A knowledge graph was constructed based on the proposed model ISPN on a real transmission and substation equipment fault dataset. The precision, recall and F1 score of the entity-relationship extraction experiments are improved by 3.9, 5.3 and 4.6 percentage points over the model, respectively. The constructed knowledge graph is stored and visualized using the Neo4j graph database, in which the structured stored fault information can provide support for subsequent fault processing and decision analysis.

Key words： Power transmission and transformation equipment knowledge graph joint extraction contrastive learning bounding regression

Received: 01 August 2024

PACS:	TM72
	TP391.1

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Yan Guangwei
	Zhang Yunxin
	Fu Zheyuan
	Jiao Runhai

Cite this article:

Yan Guangwei,Zhang Yunxin,Fu Zheyuan等. The Construction Method of Fault Knowledge Graph for Transmission and Transformation Equipment Based on Improved Set Prediction Network[J]. Transactions of China Electrotechnical Society, 2025, 40(15): 4976-4987.

URL:

https://dgjsxb.ces-transaction.com/EN/10.19595/j.cnki.1000-6753.tces.241362 OR https://dgjsxb.ces-transaction.com/EN/Y2025/V40/I15/4976

[1] 田嘉鹏, 宋辉, 陈立帆, 等. 面向知识图谱构建的设备故障文本实体识别方法[J]. 电网技术, 2022, 46(10): 3913-3922.
Tian Jiapeng, Song Hui, Chen Lifan, et al.Entity recognition approach of equipment failure text for knowledge graph construction[J]. Power System Technology, 2022, 46(10): 3913-3922.
[2] Zhu Xiangru, Li Zhixu, Wang Xiaodan, et al.Multi-modal knowledge graph construction and application: a survey[J]. IEEE Transactions on Knowledge and Data Engineering, 2024, 36(2): 715-735.
[3] 卢奥煊, 季天瑶, 李梦诗, 等. 基于数学形态学和Pearson相关系数的风电变流器开路故障识别方法[J]. 电工技术学报, 2025, 40(11): 3446-3459.
Lu Aoxuan, Ji Tianyao, Li Mengshi, et al.Identification of open-circuit fault in wind power converter based on mathematical morphology and Pearson correlation coefficient[J]. Transactions of China Electrotech-nical Society, 2025, 40(11): 3446-3459.
[4] 蒲天骄, 谈元鹏, 彭国政, 等. 电力领域知识图谱的构建与应用[J]. 电网技术, 2021, 45(6): 2080-2091.
Pu Tianjiao, Tan Yuanpeng, Peng Guozheng, et al.Construction and application of knowledge graph in the electric power field[J]. Power System Technology, 2021, 45(6): 2080-2091.
[5] 张仰森, 刘帅康, 刘洋, 等. 基于深度学习的实体关系联合抽取研究综述[J]. 电子学报, 2023, 51(4): 1093-1116.
Zhang Yangsen, Liu Shuaikang, Liu Yang, et al.Joint extraction of entities and relations based on deep learning: a survey[J]. Acta Electronica Sinica, 2023, 51(4): 1093-1116.
[6] 郭榕, 杨群, 刘绍翰, 等. 电网故障处置知识图谱构建研究与应用[J]. 电网技术, 2021, 45(6): 2092-2100.
Guo Rong, Yang Qun, Liu Shaohan, et al.Construction and application of power grid fault handing knowledge graph[J]. Power System Technology, 2021, 45(6): 2092-2100.
[7] 谢庆, 蔡扬, 谢军, 等. 基于ALBERT的电力变压器运维知识图谱构建方法与应用研究[J]. 电工技术学报, 2023, 38(1): 95-106.
Xie Qing, Cai Yang, Xie Jun, et al.Research on construction method and application of knowledge graph for power transformer operation and maintenance based on ALBERT[J]. Transactions of China Electro-technical Society, 2023, 38(1): 95-106.
[8] 何俊, 刘鹏, 聂勇, 等. 基于Seq2seq实体关系联合抽取的电力知识图谱构建[J]. 实验室研究与探索, 2022, 41(7): 1-5, 17.
He Jun, Liu Peng, Nie Yong, et al.Construction of knowledge graph of transmission regulation documents based on Seq2seq jonit extraction of entity relation[J]. Research and Exploration in Laboratory, 2022, 41(7): 1-5, 17.
[9] 束嘉伟, 杨挺, 耿毅男, 等. 面向电力知识图谱构建的重叠实体关系联合抽取方法[J]. 高电压技术, 2024, 50(11): 4912-4922.
Shu Jiawei, Yang Ting, Geng Yinan, et al.Joint extraction method for overlapping entity relationships in the construction of electric power knowledge graph[J]. High Voltage Engineering, 2024, 50(11): 4912-4922.
[10] Zhang Yingqiang, He Kejia, Xu Huifang, et al.Joint extraction of entity relations in electric power industry standard based on GCN[C]//2023 IEEE 6th International Electrical and Energy Conference (CIEEC), Hefei, China, 2023: 1249-1254.
[11] Sui Dianbo, Zeng Xiangrong, Chen Yubo, et al.Joint entity and relation extraction with set prediction networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 35(9): 12784-12795.
[12] Ziv A, Gat I, Le Lan G, et al. Masked audio generation using a single non-autoregressive transformer[J/OL]. ArXiv, 2024: 2401.04577[2024-10-30]. https://arxiv.org/abs/2401.04577v2.
[13] 汤健, 侯慧娟, 盛戈皞, 等. 变压器不平衡样本故障诊断的过采样和代价敏感算法[J]. 高压电器, 2023, 59(6): 93-102.
Tang Jian, Hou Huijuan, Sheng Gehao, et al.Oversampling and cost-sensitive algorithm for transformer fault diagnosis with unbalanced samples[J]. High Voltage Apparatus, 2023, 59(6): 93-102.
[14] 雷蕾潇, 何怡刚, 姚其新, 等. 基于变权属性矩阵的变压器零样本故障诊断技术[J]. 电工技术学报, 2024, 39(20): 6577-6590.
Lei Leixiao, He Yigang, Yao Qixin, et al.Zero-shot fault diagnosis technique of transformer based on weighted attribute matrix[J]. Transactions of China Electrotechnical Society, 2024, 39(20): 6577-6590.
[15] Wang Jiajia, Huang J X, Tu Xinhui, et al.Utilizing BERT for information retrieval: survey, applications, resources, and challenges[J]. ACM Computing Surveys, 2024, 56(7): 1-33.
[16] Tang Shunpu, Yang Qianqian, Fan Lisheng, et al.Contrastive learning-based semantic communications[J]. IEEE Transactions on Communications, 2024, 72(10): 6328-6343.
[17] Gao Tianyu, Yao Xingcheng, Chen Danqi.SimCSE: simple contrastive learning of sentence embeddings[C]//Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, Online, 2021: 6894-6910.
[18] Carion N, Massa F, Synnaeve G, et al.End-to-end object detection with transformers[C]//Computer Vision - ECCV 2020, Glasgow, UK, 2020: 213-229.
[19] 刘传洋, 吴一全, 刘景景. 无人机航拍图像中绝缘子缺陷检测的深度学习方法研究进展[J]. 电工技术学报, 2025, 40(9): 2897-2916.
Liu Chuanyang, Wu Yiquan, Liu Jingjing.Research progress of deep learning methods for insulator defect detection in UAV based aerial images[J]. Transactions of China Electrotechnical Society, 2025, 40(9): 2897-2916.
[20] Benítez-Peña S, Blanquero R, Carrizosa E, et al.Cost-sensitive probabilistic predictions for support vector machines[J]. European Journal of Operational Research, 2024, 314(1): 268-279.
[21] Asratian A S, Denley T M J, Häggkvist R. Bipartite Graphs and Their Applications[M]. Cambridge: Cambridge University Press, 1998.
[22] 万海洋, 刘文霞, 石庆鑫, 等. 城市电网灾后应急资源的集中匹配-分布调度策略[J]. 电工技术学报, 2024, 39(23): 7463-7480.
Wan Haiyang, Liu Wenxia, Shi Qingxin, et al.A post-disaster centralized matching and decentralized dispatch strategy for emergency resources in urban power network[J]. Transactions of China Electro-technical Society, 2024, 39(23): 7463-7480.
[23] 梁喻, 陈明明, 刘凡. 利用改进匈牙利算法求解旅行商问题[J]. 科学技术与工程, 2024, 24(14): 5920-5927.
Liang Yu, Chen Mingming, Liu Fan.Improvd Hungarian algorithm for solving problem of travel providers[J]. Science Technology and Engineering, 2024, 24(14): 5920-5927.
[24] Lample G, Ballesteros M, Subramanian S, et al.Neural architectures for named entity recognition[C]// Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, San Diego, California, 2016: 260-270.
[25] Wei Zhepei, Su Jianlin, Wang Yue, et al.A novel cascade binary tagging framework for relational triple extraction[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, Online, 2020: 1476-1488.
[26] Wang Yucheng, Yu Bowen, Zhang Yueyang, et al.TPLinker: single-stage joint extraction of entities and relations through token pair linking[C]//Proceedings of the 28th International Conference on Computational Linguistics, Online, 2020: 1572-1582.
[27] Shang Yuming, Huang Heyan, Mao Xianling.OneRel: joint entity and relation extraction with one module in one step[C]//Proceedings of the AAAI Conference on Artificial Intelligence, Online, 2022: 11285-11293.